Better Reads

Large-scale book recommendation engine built on 32M+ reader reviews.

This repository is source-available, not open source. Viewing for educational and reference purposes is permitted. All other use is prohibited. See LICENSE for full terms. Violations will be pursued.

---

Overview

Better Reads is a production-grade collaborative filtering system trained on one of the largest publicly available book review datasets:

• 26 million Amazon book reviews
• 6 million Goodreads reviews
• 32M+ total reader signals processed

The system generates personalized book recommendations by modeling latent reader preferences at scale.

Technical Architecture

Data Pipeline

• Distributed ingestion via AWS (S3 + EMR)
• Storage and indexing in MongoDB
• Feature engineering and batch processing with PySpark

Recommendation Engine

• Implicit ALS (Alternating Least Squares) for collaborative filtering
• Ensemble of 5 models trained on complementary vector spaces:
  • User-to-user similarity vectors
  • User-to-item interaction vectors
  • Item-to-item co-occurrence vectors
• Model training via PyTorch, Spotlight, and Implicit (Cython + OpenMP optimized)

NLP Layer

• Topic modeling on review text via NLTK
• Sentiment extraction for review quality weighting
• LLM-powered recommendation generation (see recommend.py)

Quick Start

# LLM-powered recommendations (requires vLLM, Ollama, or OpenAI API key)
python recommend.py "The Great Gatsby"
python recommend.py "Pride and Prejudice" --n 5

Technology Stack

AWS MongoDB PySpark PyTorch Spotlight Implicit Cython OpenMP NLTK Python

Related Work

This research directly informs the recommendation engine in Readify, an AI-powered interactive reading platform for schools and institutions.

---

Legal Notice

Copyright (c) 2018-2026 Clarence Stephen. All rights reserved.

This software is provided under a Source Available License. It is not open source. You may view this code for educational and reference purposes only. Commercial use, redistribution, modification, derivative works, and incorporation into other products or services are strictly prohibited without prior written authorization.

Unauthorized use will result in legal action, including DMCA takedowns, injunctive relief, and claims for damages. See LICENSE for complete terms.

For licensing inquiries or institutional partnerships: clarence@ireadifybooks.com